A critical period for second language acquisition: Evidence from 2/3 million English speakers

May combine two claims.

Percentage-point differences are unequal in meaning, larger near the extremes than near 50%, and percentage scoring creates ceiling effects that raise Type I and Type II error rates.

“Specifically, percentage points are not all of equal value, being more meaningful closer to 0% or 100% than when near 50% (Jaeger, 2008). That is, the difference between 95% and 96% is “larger” than the difference between 55% and 56%. Thus, the use of percentages artificially imposes ceiling effects, inflating both Type I and Type II error rates, particularly for interactions.”

• cites Jaeger, 2008 sourcePartly supported

  The passage on binomial variance (labelled Abstract in our copy; page unavailable) says differences near 0.5 matter less than differences near 0 or 1. The paper also describes spurious ANOVA interactions and an interaction that could be missed when proportions appear equal. It does not say that percentage scoring itself creates ceiling effects or establish that it inflates both Type I and Type II error rates; its example instead involves participants performing close to ceiling. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

May combine two claims.

Neither typical university participant pools nor services such as Amazon Mechanical Turk offer sufficient size or diversity.

“The standard undergraduate subject pools are not nearly large or diverse enough to achieve this, nor are crowdsourcing platforms like Amazon Mechanical Turk (Stewart et al., 2015).”

• cites Stewart et al., 2015 sourcePartly supported

  The Discussion (p. 7) estimates that a typical laboratory can reach about 7,300 MTurk workers—only a few times the active pool at a typical university—and warns that this may be surprisingly small. The Introduction (p. 1) says MTurk is more representative than an undergraduate sample, but the paper does not establish that MTurk lacks sufficient diversity or that either pool is too small for the citing sentence’s unspecified goal. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Neural network models can display first-language interference and surprising nonlinear behavior.

“Neural network models, which are capable of showing interference from a first language (Hernandez et al., 2005), can exhibit surprising nonlinearities (Haykin, 1999; Hernandez et al., 2005).”

• cites Haykin, 1999Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Hernandez et al., 2005 sourcePartly supported

  The Introduction describes DevLex as a neural-network model and says that, in its terms, second-language words can remain dependent on first-language forms. This supports the first-language interference part. The passages do not say that neural-network models exhibit surprising nonlinearities; a search for nonlinear behavior did not find that point in our copy. Page numbers were not provided. (LLM, conf=0.83)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Some cognitive abilities, including procedural memory, lack well-characterized developmental trajectories.

“However, the developmental trajectories of some cognitive abilities, such as procedural memory, have not been well characterized (Fuhrmann et al., 2015; Hartshorne & Germine, 2015).”

• cites Fuhrmann et al., 2015 sourceCould not verify

  The sections on cognitive control and memory and on plasticity in working memory describe developmental changes, but do not say that the trajectories of some cognitive abilities are poorly characterized. Searches for procedural memory and for poorly characterized developmental trajectories found no passage addressing that point; the working-memory section’s uncertainty concerns the effects of training across age groups instead. (LLM, conf=0.73)

  Experimental — accuracy not yet measured
• cites Hartshorne & Germine, 2015 sourcePartly supported

  The General Discussion says that questions about age-related differences between cognitive abilities remain unresolved, supporting the broad point that some trajectories are not well characterized. The paper does not address procedural memory, so it does not support that example specifically. The indexed passages provide section labels but no page numbers. (LLM, conf=0.79)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Some researchers propose that adolescence may contain several biologically driven critical periods.

“some researchers have suggested that adolescence may involve a number of different biologically-driven critical periods (Crews, He, & Hodge, 2007; Fuhrmann, Knoll, & Blakemore, 2015; see also Ghitza & Gelman, 2014).”

• cites Crews et al., 2007 sourceNot checked — no full text of the source
• cites Fuhrmann et al., 2015 sourcePartly supported

  The 'Brain development in adolescence' section says adolescence has been proposed as a sensitive period involving heightened neural plasticity, and identifies three domains in which plasticity may be heightened. The 'Models of plasticity in adolescence' section describes possible developmental profiles, but does not propose several distinct biologically driven critical periods within adolescence. Page numbers are unavailable in our copy. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites Ghitza & Gelman, 2014Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

As statistical power increased, variability in estimates dropped, making a substantial difference in correlations before and after age 16 increasingly unlikely.

“As power increased, the variability in the estimates dropped dramatically, with more highly-powered studies being increasingly unlikely to find any substantial difference in the correlations before and after 16 years old.”

May combine two claims.

Native speakers did not reach asymptotic performance until around 30 years old, though most learning occurred in the first 10-20 years.

“Inspection of Fig. 5B suggests that native speakers did not reach asymptote until around 30 years old, though most of the learning takes place in the first 10–20 years.”

May combine two claims.

The youngest native speakers in the dataset, about 7 years old, already scored very well on the quiz.

“our youngest native speakers (~7 years old) were already scoring very well on our quiz (Fig. 5B).”

May combine two claims.

Non-immersion learners showed no decline from 4 to 9 years, followed by a steep decline.

“Non-immersion learners showed similar results: From 4 years to 9 years, proficiency showed no decline (in fact it increased slightly; B = 0.01; 0.01 SDs/year), followed by a steep decline (B = −0.06; 0.07 SDs/year).”

May combine two claims.

Immersion learners with average exposure ages of 2, 5, and 8 years did not differ significantly from simultaneous bilinguals, while later learners had significantly lower curves.

“These analyses revealed that the performance curves for immersion learners with average exposure ages of 2, 5, and 8 years were not significantly different from those of simultaneous bilinguals (exposure age = 0; ps > 0.31), while the curves for later learners were significantly lower (ps < 0.01).”

Probably not a substantial claim: may not pass S3 (load-bearing).

The sample was demographically diverse.

“The sample was demographically diverse (Fig. 3).”

• AssumptionRests on an assumption: The online volunteer sample adequately represents English speakers, allowing generalization.

May combine two claims.

The best-fitting ELSD model had an R2 of 0.89 and a learning-rate change starting at 17.4 years.

“The best-fitting model (R2 = 0.89) involved a rate change beginning at 17.4 years (Fig. 4E).”

• AssumptionThe authors state this as an assumption: Syntax acquisition can be modeled as an exponential learning process with a single learning rate and an experience discount factor.
• AssumptionThe authors state this as an assumption: The learning rate is constant until a critical age and then declines according to a sigmoid function.
• AssumptionThe authors state this as an assumption: Simultaneous bilinguals, immigrants, and non-immersion learners receive different amounts of English input, captured by separate experience discount factors.

Probably not a substantial claim: may not pass S3 (load-bearing).

At least 1000 native speakers represented each of 38 languages, excluding multilinguals.

“Thirty-eight languages were represented by at least 1000 native speakers, not counting individuals who had multiple native languages.”

• AssumptionRests on an assumption: The online volunteer sample adequately represents English speakers, allowing generalization.

May combine two claims.

The ELSD model fit significantly better than a no-change model (R2 = 0.66) and a step-function model (R2 = 0.70).

“The fit was significantly better than the best fit for alternative models in which learning rate did not change (R2 = 0.66) or changed according to a step function with no further decline in the learning rate after the initial drop (R2 = 0.70).”

• AssumptionThe authors state this as an assumption: Syntax acquisition can be modeled as an exponential learning process with a single learning rate and an experience discount factor.
• AssumptionThe authors state this as an assumption: The learning rate is constant until a critical age and then declines according to a sigmoid function.
• AssumptionThe authors state this as an assumption: Simultaneous bilinguals, immigrants, and non-immersion learners receive different amounts of English input, captured by separate experience discount factors.

May combine two claims.

The same pattern appeared for hard syntactic phenomena and for easy ones usually acquired early.

“This finding held not only for “difficult” syntactic phenomena but also for “easy” syntactic phenomena that are normally mastered early in acquisition.”

• AssumptionRests on an assumption: The computational model's separation of age, exposure, and experience yields a valid estimate of underlying ability.

May combine two claims.

Girls' learning ability does not decline earlier than boys', even though girls reach puberty sooner.

“Also casting doubt on the effect of hormones is our finding that girls do not show a decline in learning ability before boys do, despite their earlier age of puberty (see Supplementary Materials).”

May combine two claims.

Ultimate-attainment differences between language groups were small and mostly unreliable.

“In fact, the differences across language groups were small (see Fig. S14) and generally not reliable.”

May combine two claims.

The study found no evidence that the learning rate slowed at different times for items mastered early (ages 7-8) versus items mastered only by older monolinguals; the best-fitting models estimated slowing at 17.3 and 18.2 years, respectively.

“We found no evidence of such a difference: In the best-fitting models of learning, the learning rate began to slow at approximately the same time for the 47 items that are mastered by the youngest monolingual English-speakers in the sample (ages 7–8) as for the 48 items that are mastered only by the older ones: 17.3 years old and 18.2 years old, respectively.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Among immersion and non-immersion learners, the largest language families were Uralic (54,664), Slavic (41,640), West Germanic (38,385), Romance (40,476), Turkic (29,816), and Chinese (15,161).

“The best-represented language families among immersion and non-immersion learners were Uralic (N = 54,664), Slavic (N = 41,640), West Germanic (N = 38,385), Romance (N = 40,476), Turkic (N = 29,816), and Chinese (N = 15,161).”

May combine two claims.

Any effect of older subjects being better at test-taking is inconsistent and small.

“As inspection of the figures indicates, any such effect is inconsistent and small.”

May combine two claims.

Most analyses supported the null of no difference between a target language and the others, and group differences were inconsistent: Romance led for age-0 starters, West Germanic for 1-5 years, and Chinese for 6-10 years.

“In most cases, analyses favored the null hypothesis (no difference between the target language and the other languages), and differences across language groups were inconsistent: among learners who began at age 0, the best-performing language group was Romance, for learners beginning at 1–5 years old, it was West Germanic, and for learners who began at 6–10 years old, it was Chinese.”

May combine two claims.

After age 10-12, attainment declines continuously with age of first exposure, with no evidence that the relationship stops at a particular age.

“After that age, we find a continuous decline in attainment as a function of age of first exposure, with no evidence that this relationship ceases after a particular age”

May combine two claims.

Evidence for learning-curve differences was somewhat stronger: simultaneous English-Chinese speakers differed from others, while Romance or West Germanic simultaneous bilinguals followed the overall pattern.

“We found slightly more evidence for differences in learning curves. In particular, simultaneous English-Chinese speakers could be distinguished from the rest, whereas simultaneous bilinguals who spoke Romance or West Germanic languages both matched the overall pattern.”

Probably not a substantial claim: may not pass S3 (load-bearing).

In the sample, Chinese and West Germanic speakers usually began immersion English earlier than Turkic or Uralic speakers (5.2 and 5.9 years vs. 13.4 and 14.8 years).

“For instance, in our dataset, speakers of Chinese and Western Germanic languages tended to start learning English in immersion settings earlier than speakers of Turkic or Uralic languages (5.2 and 5.9 years old vs. 13.4 and 14.8 years old, respectively).”

May combine two claims.

Simulations suggest prior null results were false negatives due to low power; high-power simulations suggested an optimal period ending in early or middle childhood.

“Our simulations suggest these prior findings were false negatives due to low power: Like the majority of prior studies, low-power simulations elicited largely null results, whereas high-power simulations suggested an optimal period ending in early or middle childhood (Fig. 8, right).”

May combine two claims.

The study extends the simultaneous-bilingual disadvantage to syntax, visible across the lifespan.

“We extend this finding to syntax, where it is apparent throughout the lifespan Fig. 5B).”

Probably not a substantial claim: may not pass S3 (load-bearing).

The study also finds a gender main effect, with females scoring more accurately.

“We likewise find a main effect for gender, with higher accuracy by females (see Supplementary Materials, “Gender Differences”).”

May combine two claims.

Childhood second-language learners often resemble native speakers, whereas adult-onset learners frequently retain an accent and make grammatical errors.

“People who learned a second language in childhood are difficult to distinguish from native speakers, whereas those who began in adulthood are often saddled with an accent and conspicuous grammatical errors.”

• cites Johnson & Newport, 1989 sourceNot checked — no full text of the source
• cites Lenneberg, 1967 sourceNot checked — no full text of the source
• cites Pinker, 1994Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

The causes and character of the second-language critical period remain poorly understood.

“However, neither the nature nor the causes of this “critical period” for second language acquisition are well understood.”

• cites Birdsong, 2017Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Hakuta et al., 2003 sourceNot checked — no full text of the source
• cites Johnson & Newport, 1989 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Those with inconsistent or implausible demographic answers were removed.

“We further excluded participants who gave inconsistent or implausible responses to the demographic questions”

• AssumptionRests on an assumption: Self-reported current age, first English exposure age, and time in an English-speaking country are accurate.

Classified as probably not a substantial claim.

Of 132 final items, 95 were critical: those where at least 70% of native English-speaking adults aged 18-70 in each of 13 dialects chose the same answer.

“The final set of 132 items is provided in the Supplementary Materials. Of these, 95 were critical items, defined as items for which the same response was selected by at least 70% of the native English speaking adults 18–70 years old in our full dataset in each of thirteen broadly-defined English dialects”

• AssumptionRests on an assumption: Items selected by high native-speaker agreement validly index proficiency.

Classified as probably not a substantial claim.

Understanding age-related changes in language-learning ability requires separating it from exposure age, experience length, and testing age.

“to understand how language-learning ability changes with age, we must disentangle it from age of exposure, years of experience, and age at testing.”

May combine two claims.

For non-immersion learners, the ceiling was lower but the overall pattern was the same: little difference among learners starting within the first decade, with a noticeable drop for later learners.

“For non-immersion learners, the ceiling was lower but the overall story was the same: little difference between learners who start within the first decade of life, with a ceiling that noticeably drops for later learners.”

Classified as probably not a substantial claim. May combine two claims.

The duration of the critical period, whether ability falls smoothly or sharply after it, and whether adult decline continues or levels off are all unresolved.

“We do not even know how long the critical period lasts, whether learning ability declines gradually or precipitously once it is over, or whether the ability continues to decline throughout adulthood or instead reaches a floor”

• cites Birdsong & Molis, 2001 sourceNot checked — no full text of the source
• cites Guion et al., 2000 sourceNot checked — no full text of the source
• cites Hakuta et al., 2003 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

Underlying learning ability cannot be observed directly and is hard to measure.

“Unfortunately, learning ability is a hidden variable that is difficult to measure directly.”

Classified as probably not a substantial claim.

After smoothing and filtering, the analysis retained 244,840 monolinguals, 44,412 immersion learners, and 257,998 non-immersion learners.

“leaving 244,840 monolinguals, 44,412 immersion learners, and 257,998 non-immersion learners.”

Probably not a substantial claim: may not pass S2 (checkable).

Recovering the underlying ability trajectory requires a language measure with enough sensitivity.

“Moreover, an underlying ability curve can be ascertained only if the measure of language attainment is sufficiently sensitive”

May combine two claims.

Only one previous dataset, based on US Census data, had sufficient sample size, but it used a four-point self-report of proficiency that is unlikely to be precise.

“Only one previous dataset, based on US Census data, reaches sufficient sample size (Hakuta et al., 2003; Stevens, 1999). However, this study was based on a self-report of proficiency on a four-point scale, which is unlikely to have much precision.”

• cites Hakuta et al., 2003 sourceNot checked — no full text of the source
• cites Stevens, 1999 sourceNot checked — no full text of the source

May combine two claims.

An abrupt attainment decline starting around age 10 is compatible with either gradual ability decline from infancy or a sudden learning-rate drop at 15.

“Conversely, a rapid drop in ultimate attainment beginning at age 10 could be explained by a continuous decline in learning ability beginning in infancy (Fig. 2C) or by a discontinuous drop in learning rate at 15 years old (Fig. 2D).”

Classified as probably not a substantial claim. May combine two claims.

Irregular-verb over-regularizations appear in just 0.75% of utterances and 2.5% of past-marked irregular verbs.

“over-regularizations of irregular verbs, such as runned and breaked, are among the most frequently noted errors in preschoolers’ speech (Pinker, 1999), despite occurring in only 0.75% of utterances (and on 2.5% of past-marked irregular verbs; Marcus et al., 1992).”

• cites Pinker, 1999Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Marcus et al., 1992 sourceNot checked — no full text of the source

May combine two claims.

The findings favor a sharply bounded critical period for language learning, with an offset age far later than earlier proposals.

“The results support the existence of a sharply-defined critical period for language acquisition, but the age of offset is much later than previously speculated.”

• AssumptionRests on an assumption: The computational model's separation of age, exposure, and experience yields a valid estimate of underlying ability.

May combine two claims.

Finding a 0.75% error requires substantial data; a preschooler needs 92 utterances for a better-than-even chance of one over-regularization.

“Detecting an error that occurs as little as 0.75% of the time requires a lot of data: A preschooler has to produce 92 utterances to have a better than even chance of producing an over-regularization.”

May combine two claims.

Native proficiency entails extreme accuracy; an error rate of 0.75% is conspicuous immaturity in language.

“Indeed, the concept of native proficiency entails extreme levels of accuracy. An error rate that would be considered excellent in other academic or psychological settings, such as 0.75%, represents a conspicuous immaturity in the context of language.”

• cites Pinker, 1999Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Marcus et al., 1992 sourceNot checked — no full text of the source

May combine two claims.

Monolingual reliability exceeded chance (0.66), suggesting their sparse errors were neither random nor confined to a few poor items.

“However, reliability for monolinguals was still well above chance (0.66), indicating that what few errors they made were not randomly distributed (as would be expected from mere sloppiness) nor concentrated on a few “bad” items”

• AssumptionRests on an assumption: The test items reflect syntactic knowledge, not unrelated participant traits.

Classified as probably not a substantial claim.

Even conspicuous errors require testing numerous participants across numerous items.

“Thus, to detect even “conspicuous” errors, such as childhood over-regularization, we need to test many subjects on many items.”

Probably not a substantial claim: may not pass S3 (load-bearing).

These findings are consistent with the protracted learning trajectory observed in the data.

“These findings are consistent with the protracted trajectory of learning that we observe in our data”

Probably not a substantial claim: may not pass S3 (load-bearing).

Two other methods of estimating changes in slope gave similar results.

“Two other methods of estimating changes in slope provided similar results”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

The results seem robust across multiple analyses.

“these findings appear robust and emerge in a variety of different analyses.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Ages below 7 and above 89 were treated as implausible and removed.

“we excluded participants younger than 7 and older than 89 as implausible.”

• AssumptionRests on an assumption: Self-reported current age, first English exposure age, and time in an English-speaking country are accurate.

Probably not a substantial claim: may not pass S3 (load-bearing).

About four fifths of syntax-question completers also filled out the demographic survey.

“Approximately 80% of subjects who completed the syntax questions also completed this demographic questionnaire.”

• AssumptionRests on an assumption: Self-reported current age, first English exposure age, and time in an English-speaking country are accurate.

May combine two claims.

Using diverse syntactic items gives broader coverage than many earlier studies and reduces first-language transfer effects.

“This broad approach has two advantages. First, it provides a more comprehensive assessment of syntactic phenomena than many prior studies, which focused on a smaller number of phenomena (Flege, Yeni-Komshian, & Liu, 1999; Johnson & Newport, 1989; Mayberry & Lock, 2003). Second, this diversity provides some robustness to transfer from the first language.”

• cites Flege, Yeni-Komshian, & Liu, 1999 sourceNot checked — no full text of the source
• cites Johnson & Newport, 1989 sourceNot checked — no full text of the source
• cites Mayberry & Lock, 2003 sourceNot checked — no full text of the source

• AssumptionRests on an assumption: The test items reflect syntactic knowledge, not unrelated participant traits.

May combine two claims.

Grammar-learning ability is preserved throughout childhood and declines rapidly in late adolescence.

“Taken together, the analyses above all point to a grammar-learning ability that is preserved throughout childhood and declines rapidly in late adolescence.”

• AssumptionThe authors state this as an assumption: Syntax acquisition can be modeled as an exponential learning process with a single learning rate and an experience discount factor.
• AssumptionThe authors state this as an assumption: The learning rate is constant until a critical age and then declines according to a sigmoid function.
• AssumptionThe authors state this as an assumption: Simultaneous bilinguals, immigrants, and non-immersion learners receive different amounts of English input, captured by separate experience discount factors.

Probably not a substantial claim: may not pass S2 (checkable).

The analyses indicate the findings are fairly robust, especially relative to earlier work.

“The analyses above suggest that our findings are reasonably robust, particularly in comparison to those of previous studies.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Because the quiz covered many phenomena, each type has few items, so separating item-level from type-level differences is hard.

“Unfortunately, our quiz was designed to cover a wide range of phenomena, and thus we have few items of any given type, making it difficult to distinguish differences between items and differences between item types.”

May combine two claims.

Most analyses depended on age of first exposure rather than current age, which should weaken effects of current age.

“most of our analyses did not depend on the current age of the subject but on their age at first exposure, which should weaken any effects of current age.”

Probably not a substantial claim: may not pass S3 (load-bearing).

These findings are consistent with the learning-rate results.

“These findings are consistent with our results for learning rate.”

Classified as probably not a substantial claim.

Averaging within bins is hard to apply to data subsets, as many bins contain few or no subjects.

“This is not applied easily to subsets of the data: too many bins have few or no subjects.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Written tests require literacy.

“Written tests require literacy.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Listening places high demands on speed and memory, and speech must be analyzed by non-native acoustic phonetics and phonology, which the study does not test.

“Listening places high demands on speed and memory (one can re-read but not rehear), and the speech must be analyzed by non-native acoustic phonetics and phonology, which we do not test here.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Differences between production and comprehension and between written and oral modalities can affect comparisons between native and non-native speakers.

“Certainly, differences between production and comprehension and between written and oral modalities can affect comparisons between native and non-native speakers (Bialystok & Miller, 1999).”

• cites Bialystok & Miller, 1999 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Production allows one to strategically avoid difficult and imperfectly learned words and constructions.

“Production allows one to strategically avoid difficult and imperfectly learned words and constructions.”

May combine two claims.

It cannot be concluded that differences in stimuli or population do not matter for ultimate attainment; only that studying such effects requires very large datasets.

“we cannot conclude that differences in stimuli or population do not matter for ultimate attainment, only that studying such effects requires very large datasets.”

Classified as probably not a substantial claim.

There is no computationally practical way to compare model fits across different datasets.

“In any case, we lack a computationally tractable method for comparing model fits for different datasets.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Too few subjects made most other comparisons impossible.

“Most other comparisons were not possible due to insufficiently many subjects (see Supplementary Materials).”

May combine two claims.

The findings probably do not depend on a single language or family, since participants reported over 6000 native languages or combinations.

“Our results are unlikely to be specific to any one language or language family: Participants listed more than 6000 native languages or combinations of them.”

May combine two claims.

Power analyses indicate the sample could detect only fairly large effects, so smaller differences remain possible.

“That said, power analyses suggest that we only had sufficient subjects to detect relatively large effects, meaning that we cannot rule out more subtle differences (see Supplementary Materials, under “Item Effects”).”

May combine two claims.

The learning-curve differences are small and appear to reflect somewhat faster early learning among Chinese speakers.

“However, the actual differences are subtle and seem to reflect slightly faster initial learning by the Chinese speakers (Fig. S18).”

May combine two claims.

The critical period is not explainable by early-life neuronal death or syntactic pruning, nor by adrenarche- or puberty-related hormonal shifts.

“Thus the critical period cannot be attributed to neuronal death or syntactic pruning in the first few years of life, nor to hormonal changes surrounding adrenarche or puberty (Johnson & Newport, 1989; Lenneberg, 1967; Pinker, 1994).”

• cites Johnson & Newport, 1989 sourceNot checked — no full text of the source
• cites Lenneberg, 1967 sourceNot checked — no full text of the source
• cites Pinker, 1994Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• AssumptionThe authors state this as an assumption: The present findings are reasonably robust.
• AssumptionThe authors state this as an assumption: The present results apply broadly to syntax acquisition by diverse learners.

May combine two claims.

Since the data are unevenly spread across ages and learner conditions, some conditions (e.g., monolinguals) may be over-fitted at the expense of others.

“Specifically, because our data are unevenly distributed across ages and learner conditions, we risk over-fitting certain conditions (such as monolinguals) at the expense of others.”

Probably not a substantial claim: may not pass S4 (based).

Results from any modality must reflect underlying grammatical ability at least to some degree.

“Results from any modality must reflect underlying grammatical ability at least to some degree”

Probably not a substantial claim: may not pass S3 (load-bearing).

People with different first languages produce distinctive errors in English.

“It is uncontroversial that speakers of different native languages make characteristic mistakes when speaking English (Schachter, 1990, among others)”

• cites Schachter, 1990 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Some prior studies suggest that ultimate attainment patterns differ by first language, but low power calls for caution.

“More systematically, some studies have suggested different patterns of ultimate attainment for speakers of different native languages (Bialystok & Miller, 1999), though caution is warranted given the extremely low power for such studies”

• cites Bialystok & Miller, 1999 sourceNot checked — no full text of the source

May combine two claims.

Grammar-learning ability for a new language, while weaker in adults than children, remains largely or fully intact until near adulthood.

“Crucially, the investigation of these issues—all of which have long been of interest but difficult to address—can now be guided by the finding that the ability to learn the grammar of a new language, though indeed compromised in adults compared to children, is largely or entirely preserved up to the cusp of adulthood.”

• AssumptionThe authors state this as an assumption: The present findings are reasonably robust.
• AssumptionThe authors state this as an assumption: The present results apply broadly to syntax acquisition by diverse learners.

Probably not a substantial claim: may not pass S3 (load-bearing).

The result aligns with prior work indicating that a sensitive enough test can separate even highly proficient bilinguals from monolinguals.

“This finding is consistent with some earlier work suggesting that a sufficiently sensitive test can distinguish even highly proficient bilinguals from monolinguals (Abrahamsson & Hyltenstam, 2008, 2009).”

• cites Abrahamsson & Hyltenstam, 2008 sourceNot checked — no full text of the source
• cites Abrahamsson & Hyltenstam, 2009 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Since the data are second-language performance, they do not directly test age effects on first-language learning.

“Finally, because our dataset consists of people’s performance in a second language, it does not directly address the question of how age affects the learning of a first language.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Evidence indicates that perceptual critical periods involve complex neurochemical and epigenetic promotion and braking of synaptic pruning and growth.

“Current evidence suggests that critical periods in perception involve a complex interplay of neurochemical and epigenetic promoters and brakes for both synaptic pruning and outgrowth (Werker & Hensch, 2015).”

• cites Werker & Hensch, 2015 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Earlier research indicates simultaneous bilinguals attain lower phonological proficiency than single-first-language speakers.

“First, prior work has indicated that simultaneous bilinguals do not reach the same level of proficiency in phonology as individuals with a single first language (Sebastián-Gallés et al., 2005).”

• cites Sebastián-Gallés et al., 2005 sourceNot checked — no full text of the source

May combine two claims.

Neither education nor gender appears to interact with first-exposure age, making them probably irrelevant to critical periods.

“In neither case do these main effects appear to interact with age at first exposure, and so they are unlikely to be relevant for critical periods.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Syntax learning unfolds over much longer periods and entails acquiring a complex, interconnected system.

“In contrast, syntax acquisition takes place over much longer intervals and involves learning a complex, interlocking system.”

Probably not a substantial claim: may not pass S3 (load-bearing).

That pattern is compatible with language-specific maturation.

“This observation is consistent with the possibility of language-specific maturation.”

• AssumptionThe authors state this as an assumption: The present findings are reasonably robust.
• AssumptionThe authors state this as an assumption: The present results apply broadly to syntax acquisition by diverse learners.

Probably not a substantial claim: may not pass S3 (load-bearing).

Cognitive testing has mostly targeted simple abilities measurable in one brief session, such as working memory.

“Moreover, cognitive testing has largely focused on simple abilities that can be measured in a single, short session (e.g., working memory).”